Sleep architecture and sleep patterns normally change with aging. In preclinical Alzheimer's disease (AD), the accumulation of amyloid plaques begins 10 to 20 years before any cognitive symptoms progress. Soluble amyloid-β (Aβ) is secreted during physiological synaptic activity. Since synaptic activity is correlated with sleep and awake state, a degree of Aβ fluctuates in a diurnal sleep pattern. In animal and human studies, a degree of sleep quality showed a significant correlation with brain levels of Aβ and a risk of AD. Conversely, Aβ aggregation would debilitate neuronal function in brain regions critical to sleep and wake promotion. This bidirectional relationship can be explained as positive feedback loop and associated factors that influence this relationship. In AD, the degree of sleep disturbances is much more severe compared with in the normal elderly. Further, Sundowning syndrome and a reduction of melatonin level cause a stressful neuropsychiatric symptoms and a disruption of physiological sleep rhythm, respectively. In AD patients, a correlation between sleep architectural modifications and learning performances has been reported. Moreover, executive function and emotional reactivity might be attenuated by sleep disturbances, too. However, sleep disturbance does not impact cognitive functions of all patients with AD. This could support an individual and potentially genetically determined susceptibility. Sleep disturbances have an important effect on patients and caregivers. It has a critical value to confirm and treat individuals with sleep disorders and to explore whether good quality of sleep in humans can decrease the progression of preclinical to symptomatic AD.